Pipeline Stages and Job Ordering
What You'll Learn
You'll learn how to organize GitLab CI/CD jobs into stages that control execution order and enable parallel execution for faster pipelines.
Understanding Stages
In GitLab CI/CD, stages act like floors in a building—each floor must be completed before moving to the next, but multiple rooms (jobs) on the same floor can be worked on simultaneously.
By default, GitLab provides stages like build, test, and deploy, but you can define your own. Jobs assigned to the same stage run in parallel (if multiple runners are available), while stages themselves execute sequentially.
How Stage Ordering Works
Stages are defined at the top of your .gitlab-ci.yml file and execute in the order you list them:
stages:
- build
- test
- deploy
Jobs are then assigned to stages using the stage: keyword. If a job doesn't specify a stage, it defaults to test.
compile_app:
stage: build
script:
- make build
unit_tests:
stage: test
script:
- make test
integration_tests:
stage: test
script:
- make integration
deploy_production:
stage: deploy
script:
- make deploy
In this example, compile_app runs first. Once it completes successfully, both unit_tests and integration_tests run in parallel. Only after all test jobs pass does deploy_production execute.
Why This Matters
This stage-based approach prevents wasting time on expensive deployment tasks if earlier tests fail, while maximizing speed by running independent jobs simultaneously.
Key Takeaway: Stages control the sequential flow of your pipeline, while jobs within the same stage run in parallel, giving you both order and efficiency in your CI/CD process.